Energy Subsidy Reform in Sub-Saharan Africa Experiences and Lessons

African Department 13/2 Energy Subsidy Reform in Sub-Saharan Africa Experiences and Lessons A staff team led by Trevor Alleyne I n t e r n a t i o n a l M o n e t a r y F u n d

13/2

2013 International Monetary Fund Cataloging-in-Publication Data Joint Bank-Fund Library Energy subsidy reform in Sub-Saharan Africa: experiences and lessons Washington, D.C.: International Monetary Fund, [2013] 73, 46 p.: col. ill., col. maps ; cm. At head of title: African Department. Includes bibliographical references. 1. Energy policy Africa, Sub-Saharan. 2. Electric utilities Government policy Africa, Sub-Saharan. 3. Fuel Prices Africa, Sub-Saharan. 4. Poor Energy assistance Africa, Sub-Saharan. I. International Monetary Fund. African Department. HD9502.A357 E53 2013 ISBN: 978-1-48436-654-7 ( paper) ISBN: 978-1-48437-508-2 (epub) ISBN: 978-1-48438-560-9 ( Mobipocket) ISBN: 978-1-48439-072-6 ( PDF) Disclaimer: The views expressed in this book are those of the authors and should not be reported as or attributed to the International Monetary Fund, its Executive Board, or the governments of any of its members. Please send orders to: International Monetary Fund, Publication Services P.O. Box 92780, Washington, D.C. 20090, U.S.A. Tel.: (202) 623-7430 Fax: (202) 623-7201 E-mail: publications@imf.org Internet: www.elibrary.imf.org www.imfbookstore.org

Contents Executive Summary Introduction vii xi 1. Energy Subsidies in Sub-Saharan Africa (SSA): Stylized Facts 1 Recent Developments in Fuel Pricing and Fiscal Implications 1 Electricity Subsidies and Cost Recovery Tariffs 8 Who Benefits from Energy Subsidies? 14 Energy Subsidies and Economic Efficiency 18 2. Electricity Sector Reform: Lessons and Policy Options 25 Introduction 25 Power Planning and Institutional Structure 30 Enhancing Efficiency, Reducing Costs, and Raising Revenue 31 Tariff Design, Changes, and Targeting 34 Access 35 Regional Electricity Trade 37 3. Lessons Learned from Attempts to Reform Energy Subsidies 39 Why Are Energy Subsidies So Attractive and Difficult to Remove? 39 A Strategy for Energy Subsidy Reform 42 Appendix 1. Survey of Fuel and Electricity Subsidies in Sub-Saharan Africa 53 Overview 53 Quantifying Fuel and Electricity Subsidies 57 Quantifying Contingent Liabilities 59 Conclusions 59 Appendix 2. Good Practices in Retail Fuel Price-Setting Mechanisms 61 Boxes Box 1. Methodologies and Key Concepts 2 Box 2. Energy Reforms Pay Off in Kenya and Uganda 28 Box 3. Mali: Rural Electrification Program Succeeds in Expanding Access 36 iii

CONTENTS Box 4. Increased Fiscal Space from Energy Reform and Its Uses 47 Box 5. Mitigating Measures Other Countries Experiences 49 Figures Figure 1. International Petroleum Product Prices, US$ a Liter 4 Figure 2. Fuel Price Pass-Through, Fuel Taxes, and Fiscal Cost 5 Figure 3a. Pretax Fuel Subsidies in Sub-Saharan Africa (2012) 6 Figure 3b. Fuel Subsidies in Sub-Saharan Africa (2012) on Posttax Basis 7 Figure 4. Sub-Saharan Africa: Electricity Pricing Mechanisms 9 Figure 5. Sub-Saharan Africa: Explicit Electricity Subsidies 10 Figure 6. Sub-Saharan Countries: Average Cost of Power Supply (US cents a kwh) 10 Figure 7. Cost Recovery: Average Tariffs as a Percent of Average Historical Costs 11 Figure 8. Sub-Saharan Countries: Distribution Line Losses of Power Utilities 12 Figure 9. Sub-Saharan Countries: Quasi-Fiscal Deficits of Power Utilities 12 Figure 10. Sub-Saharan Africa: Effective Residential Tarriffs by Consumption Levels 15 Figure 11. Distribution of Benefits from Fuel Subsidies in Sub-Saharan Africa 16 Figure 12. Distribution of Direct Impact of Increases in Gasoline and Kerosene Prices 17 Figure 13. The Deadweight Loss from a Fuel Subsidy 18 Figure 14. Fuel Taxes vs. Per Capita Road Sector Energy Consumption, 2003 08 19 Figure 15. Cost Recovery by Electricity Firms vs. Ease of Getting Electricity 20 Figure 16. Energy Subsidies vs. Public Spending on Education and Health 22 Figure 17. Fuel Taxes vs. CO 2 Emissions 24 Figure 18. Sub-Saharan Africa: Electricity Production Compared to Other Regions, 1975 2009 26 Figure 19. Sub-Saharan Africa: Residential Tariffs Compared to Other Regions 26 Figure 20. Sub-Saharan Africa: Cost Factor for Residential Tariffs, Compared to Other Regions 27 Figure 21. Sub-Saharan Africa: Potential Savings from Cross-Border Power Trade 37 Figure A1. Prevalence of Price-Setting Policies, June 2012 56 Figure A2. Prevalence of Explicit Subsidies, June 2012 56 Figure A3. Recent Attempts at Energy Subsidy Reform 57 Figure A4. Fuel and Electricity Subsidies in Percent of GDP 58 iv

Contents Figure A5. Estimated Contingent Liabilities in Percent of GDP 59 Figure A6. Mali: Simulations of the Impact of Alternative Pricing Mechanisms, 2006 2011 63 Tables Table 1. Sub-Saharan Africa: Trends in Quasi-Fiscal Deficits of Power Utilities 13 Table 2. Sub-Saharan Africa: Per Capita Spending by Income Quintiles 14 Table 3. Africa: Total Welfare Impact of Fuel Price Increases per Consumption Quintile 17 Appendix Table Table 1. Sub-Saharan Africa: Posttax Subsidies for Petroleum and Quasi-fiscal Deficits of Power Sector 54 Supplement Case Studies on Energy Subsidy Reform in Sub-Saharan Africa: Experiences and Lessons 1. Case Studies in Fuel Subsidy Reform 73 A. Ghana 73 B. Namibia 79 C. Niger 84 D. Nigeria 91 2. Case Studies of Electricity Subsidy Reform 98 A. Kenya 98 B. Uganda 105 v

This page intentionally left blank

Executive Summary Reforming energy (fuel and electricity) subsidies in sub-saharan Africa (SSA) is critical to ensuring future energy supply to realize Africa s growth potential. Although subsidies continue to absorb a large share of public resources, power generation and access levels in SSA remain well below those in other low-income countries. There is a link between those facts because energy subsidies create at least two set of problems. First, they are poorly targeted. The subsidies provide benefi ts to all segments of society, but the main benefi ciaries are the better off. Second, subsidies often create a disincentive for maintenance and investment in the energy sector, perpetuating energy shortages and low levels of access. Therefore, reforms are essential to make better use of budgetary resources for pro-poor and development spending and to facilitate the expansion of electricity output. But reforms are also diffi cult, because the public needs to be convinced that they will benefi t more from the reallocation of government spending to other purposes than they will lose from the subsidy removal. Reform efforts must therefore focus on putting together credible packages of measures that are then used to build support for reform. In spite of reform efforts, energy subsidies still absorb a large share of scarce public resources in SSA. According to IMF staff estimates, the fiscal cost of fuel subsidies, taking into account both direct subsidies and foregone taxes, amounted to 1.4 percent of the region s GDP in 2012 (Appendix 1). Quasi-fiscal deficits of state-owned electricity companies in SSA, defined as the difference between the actual revenue collected and the revenue required to fully recover the operating costs of production and capital depreciation, amounted to a further 1.4 percent of GDP in 2009 10 (see Appendix 1). These energy subsidies mostly benefit the better off, but their removal also would hurt the poor. Energy subsidies benefit mostly higher-income groups because they consume the most (Figure 11). Electricity subsidies are particularly regressive because connection to the electricity grid is highly skewed toward higher-income groups. Nevertheless, the welfare impact of eliminating subsidies (without compensating measures) would be significant for the poor because the share of total energy in their total household consumption is the same as the rich, although there are important differences in the types of energy products consumed across income groups (Table 3). Energy subsidies have a negative impact on economic efficiency, in particular on allocation of resources and on competitiveness and growth. Energy subsidies can lead to resource misallocation through overconsumption. They may crowd out more productive government spending, as indicated by a vii

EXECUTIVE SUMMARY negative relationship between fuel subsidies and public spending on health and education (Figure 16). More importantly, underpricing and subsidies can create a vicious cycle of underinvestment, poor maintenance, and inadequate supply, notably in the electricity sector and oil refining. In the electricity sector, persistent shortages and limited access further drive up costs, widening the wedge between tariff and cost-recovery levels. As a result, Africa s power infrastructure lags behind other developing regions, and there has been relatively little convergence toward better-equipped regions (Figure 18). According to World Bank estimates, improving electricity to the regional leader s level could increase SSA s annual potential output growth by two percentage points. Despite their drawbacks, universal energy subsidies are prevalent for a variety of reasons. An energy subsidy is a readily available mechanism, requiring very little administrative capacity, for governments to provide a highly visible benefit for important segments of the population. Other mechanisms simply may not exist. In addition, energy subsidies might be introduced by a desire to avoid the transmission of price spikes to the domestic economy, or to expand the access of the population to energy, or simply because of the difficulty of controlling the financial performance of energy companies, particularly state-owned ones. Energy subsidies are even more prevalent in oilexporting countries because of the availability of financing, the presence of lower institutional quality levels, and/or a desire to establish energy-intensive industries. Furthermore, in some countries the population expects to consume petroleum products at below international market prices as a way to share the country s oil wealth, even if refined products are imported. The longer the subsidies have existed, the more entrenched the opposition to reduce them. This is especially the case if their benefits have been capitalized, for example, by the adoption of energy intensive technologies and equipment in businesses. In addition, concerns about potential economy-wide loss of competitiveness and the impact of higher energy prices on inflation are usually raised in opposition to subsidy reform. In oil-exporting countries, the task of removing subsidies has proven even more challenging because it is difficult to convey to the public the rationale for products to be sold at their opportunity cost and not their cost of production. Case studies of SSA countries that have attempted to reduce energy subsidies (compiled in the supplement to this paper) suggest several lessons: First, transparency and public communication on the size of energy subsidies and their benefi ciaries is helpful to kick-start reform. In Nigeria, the government used the fact that fuel subsidies ($9.3 billion, or 4.1 percent of GDP in 2011) exceeded capital expenditure to call for reform. In Niger, the viii

Executive Summary realization that oil tax revenue shrank from 1 percent of GDP in 2005 to 0.3 percent of GDP in 2010 contributed to triggering reforms. Ghana undertook an independent poverty and social impact analysis in 2003 4 and made the findings public to make the costs and incidence of subsidies, along with the impact of their removal on different groups, well understood. Second, careful preparation, including public education and consultation with key stakeholders, is critical for success. In planning a reform, it is important to clearly outline the goals and objectives, identify main stakeholders and interest groups, and develop strategies to address their concerns. In Kenya, consultation with unions allowed the electricity reform process to proceed without the retrenchment of staff in the utilities. In addition, early in the reform process, the support of large consumers for tariff increases was secured only with a commitment to use extra revenues to expand electricity supply. In Namibia, the National Deregulation Task Force in 1996 examined fuel price deregulation through a broadly consultative process, culminating in a White Paper on Energy Policy in 1998. Third, a gradual phasing in and sequencing of subsidy reforms seem to work best. This is especially true if subsidies are large or have been in place for a long time. A gradual approach will allow time for energy consumers to adapt and prevent sharp price increases that could undermine support. A gradual approach would also be preferred when the available instruments for delivering mitigating measures to the most needy are less developed, and when time is needed to improve the government s track record on spending quality. In Namibia, fuel subsidies started to be scaled back only in 2001, a full three years after the adoption of a consensual white paper on deregulating energy prices. In the case of electricity, the complex nature of the reform process requires that it be gradual. In Kenya, subsidies were eliminated over the course of 7 8 years through a combination of tariff increases, improvements in collections, and reductions in technical losses. Fourth, strong institutions are needed to sustain energy subsidy reforms. In Tanzania, the establishment of a specialized regulatory entity, not only to issue licenses and technical regulations (e.g., on the quality requirements of fuel products), but also to keep the public constantly informed about (current and historical) prices and price structures and to review the proper functioning of the market (e.g., to investigate concerns about potential price collusion practices) seems to have played an important role in sustaining fuel subsidy reforms. ix

EXECUTIVE SUMMARY Fifth, durably reducing electricity subsidies involves much more than tariff increases. Breaking through the vicious cycle of underinvestment, poor maintenance, and high costs requires creating an environment conducive to seizing the considerable scope for efficiency gains. Low levels of public debt in many countries in sub-saharan Africa provide an opportunity for significant investment in cheaper sources of energy production. Regional production and distribution pools can yield significant economies of scale. Public and private energy distributors have considerable scope to reduce distribution losses and improve revenue collection rates. And a strong, knowledgeable, independent regulator can play a critical role in assessing how much subsidy removal is done by tariff adjustment versus cost containment. Finally, the credibility of the government s commitment to compensate vulnerable groups and use the savings from subsidy reform for well-targeted development interventions is essential for the success of energy subsidy reform. In the case of electricity, timing subsidy reform with improvements in power services, such as new capacity or more reliable supply, seems to raise the likelihood of success (Kenya). Kenya also maintained a lifeline electricity tariff (below costs) for households that consume less than 50 kwh a month (cross-subsidized by higher rates imposed on larger consumers) together with donor-financed subsidies to connect the poor to the electricity grid. In terms of measures to mitigate the impact of higher fuel or electricity price on the poor, conditional cash transfers are the most appropriate instrument. However, this may not be feasible in the short run because of administrative constraints. A range of actions has been introduced in practice. For example, in Niger and Ghana, the authorities introduced a subsidy for public transport to keep it affordable for the poor despite the increase in oil prices. x

Introduction The reform of energy subsidies is an important but challenging issue for sub-saharan African (SSA) countries. There is a relatively large theoretical and empirical literature on this issue. While this paper relies on that literature, too, it tailors its discussion to SSA countries to respond to the following questions: Why it is important to reduce energy subsidies? What are the difficulties involved in energy subsidy reform? How best can a subsidy reform be implemented? This paper uses various sources of information on SSA countries: quantitative assessments, surveys, and individual (but standardized) case studies. This paper is organized as follows. Chapter 1 discusses a few stylized facts about energy subsidies in SSA, their (quasi-)fiscal costs, distributional incidence, and their impact on economic efficiency. Chapter 2 focuses on policy issues linked to reforming the power sector and associated subsidies, while Chapter 3 presents a strategy for energy subsidy reform. A supplement to this paper builds on the lessons distilled from a number of case studies on energy subsidy reform Ghana, Namibia, Niger, Nigeria (fuel); and Kenya, Uganda (electricity). The paper was written by a staff team consisting of Trevor Alleyne, Christian Josz, Sukhwinder Singh, Mauricio Villafuerte, Javier Arze del Granado, Antonio David, Philippe Egoume-Bossogo, Farayi Gwenhamo, Mumtaz Hussain, Clara Mira, Anton Op de Beke, Edgardo Ruggiero, Slavi Slavov, and Geneviève Verdier. Research assistance was provided by Promise Kamanga, Brian Moon, and Douglas Shapiro. Administrative assistance was provided by Elise Brun and Edison Narvaez. A presentation of some preliminary results was made at a Ministerial Seminar on Energy Subsidies in SSA during the 2012 IMF-World Bank Annual Meetings in Tokyo. The paper was also an input into a joint FAD/AFR/MCD paper, Energy Subsidy Reform Lessons and Implications that was published in March 2013. xi

This page intentionally left blank

CHAPTER 1 Energy Subsidies in Sub-Saharan Africa (SSA): Stylized Facts Recent Developments in Fuel Pricing and Fiscal Implications 1 International oil and oil products prices rose sharply in 2003 2012 in two sequences ( Figure 1 ). They increased steadily since 2003 and more than doubled from early 2007 to mid 2008 when they peaked. They fell precipitously until the end of 2008 before rebounding strongly. This evolution has been challenging for many importing countries that saw their energy bills surge, but it has also made it increasingly difficult for many countries to resist social demands for less than full pass-through into retail fuel prices. Since end-2008, the fuel price pass-through in SSA has been lower than that of advanced economies and emerging Europe, but has been in line with pass-through in the rest of the world (Box 1). 2 Only about two-thirds of the increase in international prices was passed through to domestic prices ( Figure 2 ). From end-2008 to end-2011, when prices resumed their upward trend, the median pass-through in SSA was 66 percent. That was about the same level as in Latin America and Asia and Pacific, but well above that in the Middle East and Central Asia. Fuel price pass-through was higher than 100 percent in advanced economies and emerging Europe. Between end-2008 and end-2011, fiscal costs increased in SSA as a result of relatively low pass-through of international fuel price increases during that period. Increases in international fuel prices not fully passed through imply a 1 This section was prepared by Javier Arze del Granado, Philippe Egoume-Bossogo, Christian Josz, and Anton Op de Beke, with research assistance from Promise Kamanga and Douglas Shapiro. 2 The analysis of the dynamics of fuel price pass-through, fuel taxes, and fiscal costs in SSA between end-2008 and 2011 is based on data collected by the Fiscal Affairs Department of the IMF. Less than full pass-through implies a reduction in the tax per liter (or increase in the subsidy per liter). The change in fiscal cost is calculated by multiplying this change in tax/subsidy per liter by the change in annualized consumption between the two periods. 1

ENERGY SUBSIDY REFORM IN SUB-SAHARAN AFRICA: EXPERIENCES AND LESSONS Methodologies Fuel Subsidies Box 1. Methodologies and Key Concepts The estimation of fuel subsidies has usually relied on two methods (and its variants): (1) a price pass-through analysis; and, (2) a price benchmark analysis. The price pass-through analysis is dynamic in the sense of making inferences about the evolution of fuel subsidies or tax revenues over a certain period. An important advantage of this method is its simplicity in terms of data requirements. In fact, it only requires collecting domestic retail prices and the international fuel price for two points in time (for example, end-2008 and end-2011). By comparing the changes in domestic retail prices against the changes in international prices over that period, changes in fiscal tax/subsidy levels (be it in terms of lower fiscal revenue or budgetary outlays) can be obtained. In Figure 2, this methodology is used to highlight the fact that, in 2011, the median SSA country lost 1.6 percent of GDP in fuel tax revenue because of increased subsidization of fuel relative to end-2008. At the same time, this method has limitations. First, it assumes no changes in the cost structure of domestic fuels over time (e.g., transportation and distribution costs), although this may not be a serious problem when comparing two relatively close time periods. Second, it is quite sensitive to the choice of the starting point for the analysis. The price benchmark analysis relies on detailed cost structures to determine costrecovery fuel price benchmarks. The subsidy (tax) per liter of fuel product is obtained by subtracting the relevant benchmark from the domestic retail price. Benchmark prices are computed by adding CIF fuel import prices, national margins, and costs (e.g., transportation, distribution) and indirect taxes. There are various variants to this approach. On the one hand, pre- and posttax fuel subsidies can be obtained depending whether indirect taxes are excluded or included, respectively, from the measurement of the benchmark price. The presence of pretax subsidies (i.e., negative taxes) would clearly indicate operating losses within the supply chain and/or sale of fuel products. However, this measure may not fully reflect the true fiscal cost of the subsidies: even if the pretax subsidy is negative, thereby indicating positive revenue, those revenues may be less than if the rates stipulated in the official fuel pricing formula were applied. There are various options to compute posttax fuel subsidies, which seek to measure the fiscal cost (and sometimes other costs). In this study, the benchmark fuel taxation level was taken to be the sub-saharan Africa average of gross tax (i.e., VAT and excises) per liter. Such a benchmark focuses on the revenue potential of fuel taxation as stipulated by the tax rates in the countries fuel pricing formulas, where applicable. However, the recent IMF Board paper on energy subsidy reform (IMF, 2013) uses the national VAT rate to compute the benchmark fuel taxation level and also adds a corrective (or Pigouvian) tax to charge for externalities associated with CO 2 emissions, local pollution, and other 2

Energy Subsidies in Sub-Saharan Africa (SSA): Stylized Facts externalities such as traffic congestion and accidents. Clearly, the estimate of posttax subsidies will be very sensitive to the choice of the benchmark fuel taxation level. Electricity Subsidies In general, power utilities in sub-saharan Africa are quasi-fiscal entities. These utilities channel a variety of transfers to consumers through underpricing, uncollected electricity bills, and a number of other inefficiencies (e.g., large power distribution losses). However, the total cost of such transfers is not reflected in the budget because a large portion is implicit or involuntary (e.g., power theft). This study computes a unified measure of both explicit and implicit electricity subsidies called the quasi-fiscal deficit (QFD), which is defined as: the difference between the actual revenue charged and collected at regulated electricity prices and the revenue required to fully cover the operating costs of production and capital depreciation (Saavalainen and Joy ten Berge, 2006). The QFD is calculated as follows: QFD = Cost of underpricing of electricity + Cost of nonpayment of bills + Cost of excessive line losses where: Cost of underpricing of electricity = Q*(AC-P e ); where Q is the quantity of electricity billed to all types of consumers; AC is average cost of producing one kwh of electricity, including capital depreciation; and P e is the weighted average effective tariff per kwh that is applied by the power utility. The effective tariff rate is the price per kwh of electricity consumed at a specific consumption level when all charges variable and fixed are taken into account. Costs of nonpayment of power bills = Q* P e *(1-c); where c is the collection rate that varies between 0 percent and 100 percent. Costs of excessive line losses = Q* P e *(L-Ls); where L is actual line losses in the distribution of electricity as a percent of total consumption and Ls is the level of standard line losses assumed at 10 percent in case of sub-saharan Africa, in line with the generally held view of experts. Thus, the quasi-fiscal deficit of a power utility is measured as: QFDe = Q*(AC-P e ) + Q* P e *(1-c) + Q* P e *(L-Ls) = Q* (AC P e *[1-(1-c)-(L-Ls)]) Some Key Concepts Box 1. (continued) Price pass-through : Pass-through is defined as the absolute change in domestic retail prices divided by the absolute change in international prices, both in domestic currency. 3

ENERGY SUBSIDY REFORM IN SUB-SAHARAN AFRICA: EXPERIENCES AND LESSONS Box 1. (continued) Pass-through above (below) 100 percent implies that net fuel taxes (i.e., taxes less subsidies) are increasing (decreasing). Pass-through is calculated based on end-of-period data on domestic retail prices, international prices, and domestic currency exchange rates. For instance, pass-through from end-2008 to end-2011 is calculated as the change in the domestic retail price in this period (expressed in domestic currency) divided by the change in the international price in this period (also in domestic currency). Pretax fuel subsidy: The pretax fuel subsidy for gasoline, kerosene, and diesel is defined as the difference between an estimate of cost-recovery price (defined as CIF import price plus margins and costs) and domestic retail prices. This estimate is multiplied by fuel consumption to obtain the pretax fuel subsidy. All this information was obtained from official national sources. In computing total fuel subsidies both positive and negative values are added, hence products with positive taxes partially offset those with negative taxes (i.e., subsidies). Posttax fuel subsidy: The posttax fuel subsidy for gasoline, kerosene, and diesel is defined as the difference between an estimate of cost-recovery price (defined as CIF import price plus margins and costs) plus the SSA average of gross tax per liter and domestic retail prices. This estimate is multiplied by fuel consumption to obtain the posttax fuel subsidy. All this information was obtained from official national sources. In computing total fuel subsidies both positive and negative values are added, hence products with positive taxes partially offset those with negative taxes (i.e., subsidies). 0.8 Figure 1. International Petroleum Product Prices, US$ a Liter 0.6 Crude Oil 0.4 0.2 0 2000:Q1 2000:Q3 2001:Q1 2001:Q3 2002:Q1 2002:Q3 2003:Q1 2003:Q3 2004:Q1 2004:Q3 2005:Q1 2005:Q3 2006:Q1 2006:Q3 2007:Q1 2007:Q3 2008:Q1 2008:Q3 2009:Q1 2009:Q3 2010:Q1 2010:Q3 2011:Q1 2011:Q3 2012:Q1 Source: U.S. Energy Information Administration. 4

Energy Subsidies in Sub-Saharan Africa (SSA): Stylized Facts Figure 2. Fuel Price Pass-Through, Fuel Taxes, and Fiscal Cost 180 160 140 120 100 80 60 40 20 0 The median pass-through in SSA has been lower than in advanced economies and emerging Europe, but has been in line with most other countries in the world. Figure 2.1. Median Pass-Through, End-2008 to End-2011 (percent) 158 154 Advanced Emerging Europe 69 68 66 La n America and Caribbean Asia and Pacific Sub-Saharan Africa 41 Middle East and Central Asia 100 80 60 40 20 0 The median pass-through in SSA has been much lower in oil-exporting countries than in oil-importing countries. Figure 2.3. SSA, 1 Median Pass-through, End-2008 to End-2011 (percent) Oil exporters Oil importers Sub-Saharan Africa 5.6 83.4 66.0 1 Swaziland excluded from end-2003 to mid-2008. Equatorial Guinea, Cape Verde, Liberia, and Mauri us excluded from end-2003 to end-2008. Seychelles excluded from mid-2008 to end-2011. Comoros, Eritrea, Zimbabwe excluded for all periods. SSA had the second highest median fiscal cost due to the drop in the pass-through between 2008 and 2011. The median fiscal cost in SSA was highest in oil exporting countries. 3 2 1 0 1.8 Figure 2.2. Median Fiscal Cost, 2008 11 (percent of GDP) 3 1.6 0.7 0.5 3 2.5 2 1.5 Figure 2.4. Sub-Saharan Africa 4 Median Fiscal Cost, 2011 (percent of GDP) 3 All countries Oil-expor ng countries Oil-impor ng countries 1.6 2.7 1.5 1 0.4 0.9 1 2 0.5 3 Middle East and Central Asia Sub-Saharan Africa 2 Asia and Pacific La n America and Caribbean Advanced Emerging Europe 3 Fiscal cost is the shor all in annualized tax revenues using end-2011 tax rates rela ve to annualized tax revenues using end-2008 tax rates. 4 Comoros, Eritrea, Seychelles, and Zimbabwe are omi ed. 0 Sources: IMF World Economic Outlook and staff calculations. loss of tax revenue and/or increased subsidies. The median increase in fiscal cost was 1.6 percent of GDP in SSA and was second only to that of the Middle East and Central Asia ( Figure 2.2 ). These two regions experienced losses of more than twice those recorded by Asia and Pacific and Latin America and the Caribbean. There is a clear difference in the pass-through behavior of oil exporters and oil importers in SSA ( Figure 2.3 ). Looking at the period after the price shock (i.e., end-2008 to end-2011), the median pass-through for oil exporters was much lower than the median for oil importers. Oil exporters found it harder to pass through changes in international oil prices to consumers, who may consider low fuel prices the most convenient way to share in the oil wealth of their countries. As a result, the increase in fiscal cost in SSA oil-exporting countries was almost twice as high as in SSA oil-importing countries ( Figure 2.4 ). This reflects both a lower pass-through and higher fuel consumption in oil-exporting countries. 5

ENERGY SUBSIDY REFORM IN SUB-SAHARAN AFRICA: EXPERIENCES AND LESSONS Figure 3a. Pretax Fuel Subsidies in Sub-Saharan Africa (2012) (percent of GDP) 3.0 2.0 1.0 0.0 1.0 2.0 3.0 4.0 Oil exporters 0.7 0.8 0.9 0.9 0.3 0.6 0.4 2.5 2.0 3.5 3.5 3.4 Oil importers Median of oil exporters Median of oil importers 0.5 0.7 1.0 1.0 1.0 1.8 1.8 1.7 1.7 1.6 1.6 1.6 1.6 1.5 1.5 1.5 1.4 1.3 1.2 1.2 2.3 2.3 2.2 2.1 2.1 2.1 2.5 2.6 0.4 0.5 0.1 5.0 6.0 5.3 Côte d'ivoire Chad Gabon Equatorial Guinea Median oil exporters Congo, Rep. of Cameroon Nigeria Angola Malawi Guinea Namibia Senegal Benin São Tomé and Príncipe Mali Swaziland Mauri us Seychelles Zambia Burkina Faso Gambia, The Comoros Kenya South Africa Cape Verde Median oil importers Uganda Tanzania Congo, Dem. Rep. Togo Central African Rep. Burundi Madagascar Rwanda Lesotho Niger Botswana Mozambique Guinea-Bissau Sierra Leone Zimbabwe Ethiopia Ghana Sources: Authorities data and IMF staff estimates. Note. Negative values represent a tax. The pre tax data for gasoline, kerosene, and diesel in each country are calculated as the difference between an estimate of cost-recovery price (defined as: CIF import price plus national margins and costs ) and domestic retail prices In addition to looking at the dynamic behavior of fuel prices between end- 2008 and end-2011 to calculate the change in fuel taxation/subsidization, estimates of the absolute size of fuel subsidies at end-2012 were also calculated. Based on a detailed survey conducted for SSA countries, the fuel subsidies for gasoline, kerosene, and diesel were estimated based on the difference between an estimate of cost-recovery price and domestic retail prices. This estimate was multiplied by fuel consumption to obtain the fuel subsidy. Two alternative measures were computed. The pretax subsidy compares a cost-recovery price that includes the CIF import price plus national margins and costs with the retail price. The posttax subsidy compares an adjusted cost-recovery price (i.e., the CIF import price plus national margins and costs plus a measure of gross taxes per liter) with the retail price. For this paper, the SSA average gross tax per liter was used, but clearly other formulations could be justified (e.g., in IMF [2013], posttax subsidies were calculated using an adjusted cost-recovery price that includes 6

Energy Subsidies in Sub-Saharan Africa (SSA): Stylized Facts Figure 3b. Fuel Subsidies in Sub-Saharan Africa (2012) on Posttax Basis (percent of GDP) 5.0 4.0 3.0 Oil exporters 2.8 3.4 4.0 Oil importers Median of oil exporters Median of oil importers 3.2 3.3 2.0 1.0 0.0 1.0 2.0 2.1 1.9 1.6 1.7 1.2 0.7 0.2 0.1 0.0 0.0 0.7 0.6 0.5 1.1 1.0 1.6 1.8 1.2 1.2 0.1 0.1 0.2 0.2 0.2 0.2 0.4 0.4 0.5 0.5 0.6 0.7 0.7 0.7 0.8 0.8 0.8 0.9 0.9 3.0 2.5 Côte d'ivoire Equatorial Guinea Gabon Chad Median oil exporters Congo, Rep. of Cameroon Nigeria Angola Malawi Senegal Mauri us Seychelles Namibia Swaziland South Africa Central African Rep. Rwanda Mozambique Tanzania Guinea-Bissau Zambia Uganda Burkina Faso Cape Verde Niger Median oil importers Mali Guinea Comoros Madagascar Burundi Lesotho Kenya Congo, Dem. Rep. Togo Botswana Benin Sierra Leone Gambia, The Ethiopia São Tomé and Príncipe Ghana Zimbabwe Sources: Authorities data and IMF staff estimates. Note: Negative values represent a tax. The post tax data for gasoline, kerosene and diesel in each country are calculated as the difference between an estimate of cost-recovery price (defined as: CIF import price plus national margins and costs plus the SSA average gross tax per liter) and domestic retail prices. the cost of externalities, such as CO 2 emissions and traffic congestion). Although the pretax subsidy reflects the more common understanding of a subsidy, the posttax subsidy aims to measure the fiscal cost or unutilized fiscal space. Most SSA countries do not have pretax subsidies on fuel ( Figure 3a ). In other words, the retail price of fuel products is typically greater than the cost recovery price. However, while only 10 percent of oil importers have pretax subsidies, almost all of the oil exporters do and the median cost of the subsidies for this group is 0.8 percent of GDP. Posttax fuel subsidies are significantly higher and more widespread across the region ( Figure 3b ). At 1.9 percent of GDP, these subsidies were almost five times higher in oil-exporting countries than in SSA oil importing countries (0.4 percent of GDP). 7

ENERGY SUBSIDY REFORM IN SUB-SAHARAN AFRICA: EXPERIENCES AND LESSONS Electricity Subsidies and Cost Recovery Tariffs 3 This subsection gives estimates of fiscal and quasi-fiscal costs in the power sector in SSA countries and analyzes the factors that underlie these costs. A power utility company generates hidden costs when its realized revenue is less than the revenue it would collect were it operated with cost recovery tariffs based on efficient operations (i.e., operations with normal line losses and full collection of bills). In the last few decades, power companies in SSA tended to experience substantial hidden costs, which in turn constrained their ability to invest in new power capacity, to expand access, and to improve service quality. As a result, per capita installed generation capacity in SSA (excluding South Africa) is about one-third of that in South Asia and one-tenth of that in Latin America (Eberhard and Shkaratan, 2012). Similarly, per capita consumption of electricity in SSA (excluding South Africa) is merely 10 kwh a month, in contrast to about 100 kwh in developing countries and 1000 kwh in high-income countries. Most countries in SSA have highly regulated electricity markets. A survey on SSA countries (Appendix 1) suggests that most countries implement some form of administered pricing for electricity, most frequently ad hoc nonautomatic price setting schemes ( Figure 4 ). Even in countries with de jure pricing policies based on an automatic formula, these automatic mechanisms are frequently suspended or intervened. Most electricity utilities are state owned, and it appears that policymakers are reluctant to adopt market-based pricing policies, partly because of concerns related to access, affordability, and institutional capacity. Subsidies for electricity services are common in SSA. A majority of countries have explicit subsidies for electricity ( Figure 5 ). Despite shortcomings in terms of data availability, it is clear that explicit subsidies are substantial. We estimate that direct power subsidies average 0.4 percent of GDP for SSA, but can reach up to 0.8 percent (Mali). In addition, there has been a build-up of arrears by state-owned power utilities (on average 0.6 percent of GDP) and debt accumulation (on average 1.5 percent of GDP). Factors Contributing to the Under-Recovery of Power Costs Excluding South Africa, the average cost of supplying one kwh in sub- Saharan African countries is the highest among developing countries 3 Prepared by Mumtaz Hussain. 8

Energy Subsidies in Sub-Saharan Africa (SSA): Stylized Facts Figure 4. Sub-Saharan Africa: Electricity Pricing Mechanisms Pricing Liberalized Ad hoc Formula Source: Survey of IMF country teams for SSA countries (April 2012). (Eberhard and Shkaratan, 2012). Using the latest annual data for 2008 10, the average cost of electricity in SSA was about US$0.15 a kwh. The average cost of power was even higher in countries that rely primarily on thermal generation US$0.21 a kwh ( Figure 6 ). Besides inefficiencies in power companies, high use of costly emergency power generation (e.g., in Uganda until early 2012), low economies of scale in generation, and limited regional integration also contributed to these high unit costs. Effective power tariffs are generally set well below the historical average cost of supplying electricity ( Figure 7 ). 4 Despite residential tariffs in SSA countries being much higher than in other regions of the world (Briceño-Garmendia and Shkaratan, 2011), they cover, on average, only about 70 percent of the cost of power (based on data for the latest year in 2005 09). 4 The effective tariff rate is the price per kwh of electricity consumed at a specific consumption level when all charges variable and fixed are taken into account (Briceño-Garmendia and Shkaratan, 2011). 9

ENERGY SUBSIDY REFORM IN SUB-SAHARAN AFRICA: EXPERIENCES AND LESSONS Figure 5. Sub-Saharan Africa: Explicit Electricity Subsidies Subsidies Not available Subsidies No subsidies Source: Survey of IMF country teams for SSA countries (April 2012). Figure 6. Sub-Saharan Africa Countries: Average Cost of Power Generation (US cents a kwh) 25 20 15 10 5 0 Primarily hydro genera on Primarily thermal genera on Source: The World Bank s AICD database and various country reports. Data is for 2008-10 (latest year available). South Africa is excluded. 10

Energy Subsidies in Sub-Saharan Africa (SSA): Stylized Facts Figure 7. Cost Recovery: Average Tariffs as a Percent of Average Historical Costs 160 140 120 100 80 60 40 20 0 Kenya Niger Sierra Leone Namibia South Africa Congo, D. Rep Burkina Faso Malawi Rwanda Botswana Côte d'ivoire Cape Verde Congo, Rep. Source: Briceño-Garmendia and Shkaratan (2011) and various country reports from the World Bank. Senegal Madagascar SSA average Benin Mozambique Uganda Ghana Zimbabwe Lesotho Mali Cameroon Tanzania Zambia Ethiopia Nigeria Angola In addition, power utilities tend to be subject to high line losses in some cases half of power injected into the distribution system is lost and undercollection problems. On average, distribution losses (the amount of electricity injected into the distribution network that could not be billed) are around 25 percent well above the international norm of 10 percent ( Figure 8 ). Similarly, the average collection rate was around 85 percent. The costly power supply relative to per capita income in SSA has contributed to power theft and nonpayment of bills. Evidence from household surveys indicates that as much as 60 percent of poorer households do not pay their electricity bills (Briceño-Garmendia and Shkaratan, 2011). Power Sector s Quasi-Fiscal Deficits in SSA Countries Quasi-fiscal deficits of power utilities in SSA countries are large in terms of GDP ( Figure 9 ). Using the latest available data for 2008 10, the median quasifiscal deficit (QFD) was about 1.7 percent of 2009 GDP. However, there are large variations in QFDs across countries: from about 11 percent of GDP in Zimbabwe to less than 0.5 percent of GDP in Botswana and Chad. Also, a number of countries have managed to reduce deficits (e.g., Kenya) while others have experienced increased QFDs (partly due to increased reliance on emergency power generation). Kenya implemented a number of reforms in 11

ENERGY SUBSIDY REFORM IN SUB-SAHARAN AFRICA: EXPERIENCES AND LESSONS 50 Figure 8. Sub-Saharan African Countries: Distribution Line Losses of Power Utilities (Losses in percent of power supplied) 40 30 20 10 0 Angola Sierra Leone Congo, D. R. Congo Uganda Cameroon Chad Mozambique Cape Verde Ghana SSA average Madagascar Malawi Rwanda Senegal Ethiopia Mali Nigeria Lesotho Kenya Benin Niger Côte d'ivoire Burkina Faso Tanzania Namibia Zambia Zimbabwe Bostwana South Africa Source: Eberhard and others (2008) and various country reports from the World Bank. 8 7 6 5 4 3 2 1 0 Mozambique Sierra Leone Figure 9. Sub-Saharan African Countries: Quasi-Fiscal Deficits of Power Utilities (QFD in percent of 2009 GDP) 1 Zambia Cape Verde Congo, D. R. Cameroon Senegal Mali Ghana Côte d'ivoire Uganda Malawi SSA average Source: IMF staff calculations based on data from the IMF, the World Bank, and International Energy Agency. 1 Zimbabwe, which had QFD of 11 percent of GDP in 2009, is excluded from the calculation of average. Benin Tanzania Ethiopia Nigeria Madagascar Angola Burkina Faso Lesotho Congo Kenya Namibia Niger Rwanda South Africa Bostwana Chad 12

Energy Subsidies in Sub-Saharan Africa (SSA): Stylized Facts Table 1. Sub-Saharan Africa: Trends in Quasi-Fiscal Deficits of Power Utilities (Percent of GDP, averages unless otherwise noted) 2005 06 2009 10 Quasi-fiscal deficit generated by: Underpricing 1.1 0.8 Distribution line losses 0.7 0.8 Undercollection of bills 0.2 0.2 Total quasi-fiscal deificit (excluding Zimbabwe) 1.9 1.8 Total quasi-fiscal deificit (median) 1.7 1.7 Source: IMF staff calculations based on data from country authorities, the IMF, the World Bank, and International Energy Agency. its power sector during the last decade that reduced the QFD by about 0.7 percent of GDP. In any case, these estimates of total subsidy are about three times as large as the levels reported as direct fiscal transfers in the AFR survey (Appendix 1). SSA countries have made little progress in reducing QFDs ( Table 1 ). The median power sector s QFD has remained unchanged between 2005 and 2010. A slight reduction in underpricing was mostly offset by increased distribution losses. Affordability vs. Cost Recovery: Is There Room for Raising Power Tariffs to Cost Recovery Levels? Power sector reforms to enhance efficiency and reduce losses should help reduce substantially the QFD of power utilities. Average residential tariffs in SSA are already higher (in some cases twice as much) than in other regions of the world, while average per capita incomes in Africa are substantially lower. Therefore, it could be argued that tariff policy is not an effective tool to reduce QFDs because further tariff hikes only lead to lower collection rates and increased distribution losses (e.g., theft). This argument has some truth to it and points to the need to address these operational inefficiencies as part of any credible subsidy reform strategy (this is discussed in detail in Chapter 2). In fact, long-run marginal costs estimated by the World Bank are about 12 percent less than historical costs and as much as 50 percent less in some cases (e.g., Malawi, Cameroon, Botswana, Tanzania) (Briceño-Garmendia and Shkaratan, 2011). Nevertheless, cost-recovery tariffs can be achieved when combined with better services from power utilities. It is important to note that households and firms spend considerable amounts to deal with intermittent power 13

ENERGY SUBSIDY REFORM IN SUB-SAHARAN AFRICA: EXPERIENCES AND LESSONS supply and shortages (e.g., purchase and operation of petroleum-powered generators). The costs of own generation (by firms) is estimated in the range of US$0.3 US$0.7 a kwh about three to four times as high as the price of electricity from the public grid (Foster and Steinbuks, 2008). These costs are even higher for households. Who Benefits from Energy Subsidies? 5 Fuel and electricity consumption in SSA countries is highly skewed toward higher income households. Available data show that patterns of fuel and energy consumption across households in various income quintiles vary significantly ( Table 2 ). Household survey evidence from nine African countries (Arze del Granado, Coady, and Gillingham, 2012) suggests that poorer households consume directly a much smaller share of the total fuel and electricity supplied. In fact, households in the richest quintile spent on per capita terms close to 20 times more on fuel and electricity than households in the poorest quintile (kerosene is the only exception, with broadly evenly distributed consumption across households). Beside relatively higher incomes, better access to energy resources (particularly electricity in urban areas) contributes to the higher fuel consumption of richer households. Table 2. Sub-Saharan African Countries: Per Capita Spending by Household Income Groups (PPP values in 2005 U.S. dollars, sample averages) Q1 (Poorest) Q2 Q3 Q4 Q5 (Richest) Spending on diesel fuel Spending on gasoline Spending on kerosene Spending on electricity $ amount ratio of Q5 to Q1 $ amount ratio of Q5 to Q1 $ amount ratio of Q5 to Q1 $ amount ratio of Q5 to Q1 0.3 0.5 0.9 1.3 6.9 20.0 0.1 0.4 0.9 1.7 3.0 27.4 1.5 1.9 2.0 2.1 2.7 1.8 0.5 1.3 2.2 3.9 8.9 17.0 Countries in the sample are (survey year in brackets): Cameroon (2007), Côte d Ivoire (2008), Ethiopia (2004), Ghana (2005), Mozambique (2009), Rwanda (2005), Senegal (2005), Uganda (2010), and Zambia (2010). Source: World Bank (2012) Africa Pulse Database. 5 Prepared by Mumtaz Hussain and Clara Mira. This section draws on Arze del Granado, Coady, and Gillingham (2012) and World Bank (2012). 14

Energy Subsidies in Sub-Saharan Africa (SSA): Stylized Facts Figure 10. Sub-Saharan Africa: Effective Residential Tariffs by Consumption Levels (US cents per kwh) 16 14 12 10 8 6 4 2 0 50 kwh 75 kwh 100 kwh 150 kwh 200 kwh 300 kwh 400 kwh 450 kwh 500 kwh 900 kwh Source: Briceño-Garmendia and Shkaratan (2011). Furthermore, differences in effective tariffs across various electricity consumption levels are small. About two-thirds of sub-saharan African countries use increasing block tariffs (IBTs) (Briceño-Garmendia and Shkaratan, 2011). However, the progressivity of tariffs is limited in most countries partly because of relatively large fixed monthly charges. This results in rather modest differences in effective tariffs at vastly different levels of electricity consumption by households ( Figure 10 ). In this context, it is not surprising to find that fuel and electricity subsidies tend to benefit the better off. Because the richer households have higher consumption levels of fuel and electricity than the lower-income households, they capture the majority of the funds allocated to universal subsidies such subsidies are per unit of fuel or electricity consumption regardless of consumers income levels. In sub-saharan Africa, on average, the households in the top consumption quintile capture about 45 percent of fuel subsidies, while the poorer segments of the population (the bottom 40 percent of households) receive about 20 percent of the subsidy benefit ( Figure 11 ). If protecting poor and vulnerable groups is a key policy objective, universal subsidy schemes do not do a good job. The evidence suggests that providing 1 dollar of relief to the poorest 40 percent of the population under the universal subsidy policy requires the government to spend 5 dollars, of which about half would accrue to the richest quintile. However, a subsidy reform implying an increase in energy prices would still have a sizable impact on the poorest segments of the population. For 15

ENERGY SUBSIDY REFORM IN SUB-SAHARAN AFRICA: EXPERIENCES AND LESSONS Figure 11. Sub-Saharan Africa: Distribution of Benefits from Fuel Subsidies Percent of total subsidy 7.8% 44.2% 21.2% 11.5% 15.3% Poorest quin le Quin le 2 Quin le 3 Quin le 4 Richest quin le Source: Arze del Granado, Coady, and Gillingham (2010). example, an increase of $0.25 a liter in fuel prices in SSA countries would reduce, on average, the 40 percent poorest households real income by 5.7 percent ( Table 3 ). Over half of this purchasing power loss would occur through the indirect effect pass-through of higher fuel prices into food and transportation costs reflecting the importance of fuel as an intermediate input in the production process. 6 Such an impact might be even larger if distribution of electricity spending is adjusted for the disparity in the access to electricity. Electricity in SSA countries is skewed to richer households among the poorest 40 percent of households, this access rate is below 10 percent, whereas it rises to close to 80 percent for the richest household quintile (Eberhard and Shkaratan, 2012). When corrected for this disparity in access, the cost of electricity to low-income households (having access to the grid) rises substantially. For example, an analysis for Burkina Faso undertaken by Arze del Granado, Coady, and Gillingham (2012) suggests that the poorest 40 percent of the population with electricity provision devote, on average, 4.4 percent of their budgets to electricity consumption (rather than the 0.4 percent implied by an analysis including all households regardless of access). 7 Although the overall impact of a fuel price increase looks similar across income groups, there would be significant variation in the distribution of the 6 Indirect effects can be calculated through price shifting models, assuming that increases in fuel costs are fully passed to domestic prices (Arze del Granado, Coady, and Gillingham, 2012). 7 Household surveys also indicate that as much as 60 percent of households at the bottom with service do not pay their electricity bills, compared with about 20 percent of those in the highest consumption quintile (Briceño- Garmendia and Shkaratan, 2011). 16